Perceptions to Beliefs: Exploring Precursory Inferences for Theory of Mind in Large Language Models

TL;DR

This paper investigates the precursors to theory of mind in large language models, introducing new datasets and a method that improve models' understanding of beliefs, especially in false belief situations.

Contribution

The paper introduces two datasets for evaluating perception and perception-to-belief inference in LLMs and proposes PercepToM, a method that enhances ToM performance in these models.

Findings

LLMs perform well in perception inference

Limited in perception-to-belief inference, especially inhibitory control

PercepToM significantly improves false belief understanding

Abstract

While humans naturally develop theory of mind (ToM), the capability to understand other people's mental states and beliefs, state-of-the-art large language models (LLMs) underperform on simple ToM benchmarks. We posit that we can extend our understanding of LLMs' ToM abilities by evaluating key human ToM precursors$-$perception inference and perception-to-belief inference$-$in LLMs. We introduce two datasets, Percept-ToMi and Percept-FANToM, to evaluate these precursory inferences for ToM in LLMs by annotating characters' perceptions on ToMi and FANToM, respectively. Our evaluation of eight state-of-the-art LLMs reveals that the models generally perform well in perception inference while exhibiting limited capability in perception-to-belief inference (e.g., lack of inhibitory control). Based on these results, we present PercepToM, a novel ToM method leveraging LLMs' strong perception inference capability while supplementing their limited perception-to-belief inference. Experimental results demonstrate that PercepToM significantly enhances LLM's performance, especially in false belief scenarios.